Abstract This project will develop a real-time neuroimaging and behavioral method by which individuals with addiction may use brain- and keypress-based feedback to modulate their responses to drug cues, and alter brain and behavior signatures of addiction. Drug addiction can be characterized by a restructuring of the reward system to short-term rapid rewards, decreased assessment of long-term consequences, impulsivity, and a narrowing of the behavioral repertoire (i.e., restriction in the set of things toward which one shows preference). To deal with these symptoms of addiction, one therapeutic approach has been to develop urge minimization techniques through adaptations of behavioral therapy. These therapy techniques have traditionally focused on psychosocial measures of addiction that can be difficult to quantify or alter objectively. Recent advances in addiction neuroscience have revealed robust alterations of reward/aversion circuitry along with experimental psychology variables in addiction that may be amenable to use with behavioral techniques. Specifically, structural and functional MRI measures of prefrontal cortex, point to localized differences in addicts that can be connected with a keypress-based phenotype of relative preference that quantifies their restricted behavioral repertoire. Technical and computational advances suggest that these neuroscience measures might be harnessed for behavioral therapy. Toward this end, this project lays out a set of milestones as an R21 proposal to develop (a) a technical work-flow for high SNR, quantitative BOLD activation measurement based on single and double voxel spectroscopy; (b) real-time signal analysis and display software for feedback to individuals based on behavior and BOLD signal; (c) integration of platforms for imaging, experimental psychology, and feedback with subjects in the scanner. Following successful attainment of these milestones, a further set of milestones are organized as an R33 proposal to demonstrate: (d) that these technologies can be applied to monitor modulation of brain signal and relative preference phenotype by subjects, (e) integration of modulation of brain signal and keypress responses during realistic testing conditions, and (f) the application of these techniques to cocaine dependent subjects. We foresee these proof-of-concept milestones to be prerequisites for the future use of such a method in therapeutic trials.